Delivering Lync 2013 Real-Time Communications over Wi-FiPublished: January 25, 2013

Author: Peter Schmatz

Technical Reviewers: Amer Hassan, Pascal Menezes, Victoria Poncini

Editor: June Rugh

Abstract: This white paper describes how Lync 2013 communications software can be successfully implemented over wireless local area networks (Wi-Fi). The Lync 2013 client has been validated on multiple platforms with voice and video (real-time media) over wireless networks, and user experiences and best practices are summarized here. A list of issues and mitigations will help you to ensure a high-quality voice and video Lync experience for all users. To optimize the wireless infrastructure, particularly for real-time media traffic, you’ll find details regarding Wi-Fi (WLAN) technology, configuration settings, and optimization. In addition, this paper provides deployment recommendations and evaluates typical enterprise, public hotspot, and home Wi-Fi deployments for real-time bidirectional media quality.

This document is provided “as-is”. Information and views expressed in this document, including URL and other Internet Web site references, may change without notice.

Some examples depicted herein are provided for illustration only and are fictitious. No real association or connection is intended or should be inferred.

This document does not provide you with any legal rights to any intellectual property in any Microsoft product. You may copy and use this document for your internal, reference purposes.

Copyright © 2012 Microsoft Corporation. All rights reserved.

Table of Contents1 Overview............................................................................................................................................... 5

2 Introduction........................................................................................................................................... 5

2.1 Glossary......................................................................................................................................... 53 Usage Scenarios................................................................................................................................... 6

3.1 Enterprise Wi-Fi.............................................................................................................................. 63.1.1 Enterprise voice/video mobility scenarios................................................................................63.1.2 Personal devices and enterprise Wi-Fi.....................................................................................7

3.2 Home Wi-Fi.................................................................................................................................... 73.3 Public Wi-Fi hotspots...................................................................................................................... 7

4 Issues That Affect Wi-Fi Performance for Real-Time Application.........................................................8

4.1 General issues............................................................................................................................... 84.1.1 Wireless NIC drivers................................................................................................................84.1.2 Wireless NIC chipsets and hardware.......................................................................................84.1.3 Legacy interoperability issues..................................................................................................9

4.2 Issues in enterprise Wi-Fi deployments..........................................................................................94.3 Issues in public Wi-Fi hotspots.......................................................................................................94.4 Issues in home Wi-Fi deployments...............................................................................................10

5 . Wi-Fi Deployment Recommendations...............................................................................................10

5.1 Enterprise Wi-Fi............................................................................................................................ 105.1.1 Mixed or Wi-Fi-only enterprise deployment............................................................................105.1.2 Device types in enterprise Wi-Fi deployments.......................................................................105.1.3 Usage patterns in enterprise Wi-Fi.........................................................................................105.1.4 Legacy devices in enterprise Wi-Fi.........................................................................................115.1.5 WMM QoS and WMM-Power Save support...........................................................................115.1.6 Enterprise WLAN controller....................................................................................................115.1.7 Recommended Wi-Fi feature support....................................................................................125.1.8 Recommended Wi-Fi configuration settings...........................................................................125.1.9 Wi-Fi policies..........................................................................................................................125.1.10 Wireless network card (network adapter) recommendations.............................................125.1.11 Enterprise voice certifications............................................................................................13

5.2 Home Wi-Fi.................................................................................................................................. 135.2.1 Home Wi-Fi deployment recommendations...........................................................................13

5.3 Public hotspot Wi-Fi.....................................................................................................................145.3.1 Basic hotspot Wi-Fi configurations.........................................................................................14

1 Appendix A: Wi-Fi Standards..............................................................................................................15

1.1 IEEE 802.11a/b/g/n Wi-Fi standards............................................................................................151.2 802.11a (legacy standards)..........................................................................................................151.3 802.11b (legacy standards)..........................................................................................................151.4 802.11g (current standards).........................................................................................................161.5 802.11n (current standards).........................................................................................................16

1.5.1 802.11n MIMO........................................................................................................................ 16

1.5.2 802.11n Spatial diversity........................................................................................................171.5.3 Physical layer access by using a 40-MHz wide band.............................................................171.5.4 Wi-Fi Certified 802.11 n devices.............................................................................................17

1.6 802.22 and 802.11af Wireless Regional Network (future standards)............................................181.7 802.11ac Gigabit Wi-Fi (future standards)....................................................................................18

2 Appendix B: Wi-Fi Frequencies...........................................................................................................19

2.1 2.4-GHz band............................................................................................................................... 192.1.1 Coexistence of 11b/g/n in 2.4 GHz.........................................................................................19

2.2 5 GHz band.................................................................................................................................. 202.3 Quality of Service and Wi-Fi Multimedia (WMM)..........................................................................20

2.3.1 WMM prerequisites................................................................................................................202.3.2 WMM-Power Save (APSD)....................................................................................................21

3 Appendix C: Wi-Fi Security.................................................................................................................21

3.1 Wi-Fi security and real-time media workloads..............................................................................213.1.1 Pairwise Master Key Security Association (PMKSA).............................................................223.1.2 Opportunistic Pairwise Master Key (PMK) Caching...............................................................223.1.3 802.11r fast BSS transition.....................................................................................................22

4 Appendix D: Wi-Fi WAP Handover and Resource Management........................................................23

4.1 Background WAP scanning..........................................................................................................234.2 802.11k Radio resource management..........................................................................................234.3 Other industry-standard fast handover solutions..........................................................................234.4 802.11v Wireless network management.......................................................................................244.5 WMM-Power Save and DTIM interval..........................................................................................24

1 OverviewMicrosoft® Lync® 2013 communications software on multiple platforms and device types is now validated with voice and video (real-time media) workloads over wireless local area networks (Wi-Fi networks). The guidance and documentation provided here also applies to Lync 2010 deployments and earlier releases. However, full Wi-Fi support will be available only with a Lync 2013 client. Changes in the Lync media stack available with Lync 2013 will improve overall performance on high-loss, high-jitter wired or wireless networks.

To optimize the wireless infrastructure, in particular for real-time media traffic, this guide provides details regarding Wi-Fi (WLAN) technology, configuration settings, and optimization.

In addition, this guide provides deployment recommendations and evaluates typical enterprise, public hotspot, and home Wi-Fi deployments for real-time bidirectional media quality.

2 Introduction Wi-Fi connectivity is ubiquitous, and users expect to find Wi-Fi access in the office, at home, in public hotspots, in hotels, and on the road. However, the Quality of Service (QoS) in Wi-Fi varies widely, ranging from the ability to support basic email sync and web traffic, to the ability to support of a full range of multimedia applications and real-time communications.

This deployment guide discusses different usage scenarios of wireless connected devices, coverage, capacity, and QoS aspects, specifically for delivering real-time communication workloads, such as audio, video, and application sharing. You’ll also find deployment recommendations for enterprise, hotspot, and home Wi-Fi deployments, and a full discussion of issues and mitigations. Finally, the extensive appendix provides a detailed overview of legacy, current, and future Wi-Fi standards, Wi-Fi frequencies, Wi-Fi security, and Wi-Fi WAP (Wireless Access Point) handover and resource management.

By successfully deploying Lync 2013 over wireless networks—optimizing your wireless infrastructure for real-time media traffic, prioritizing types of usage, preparing for specific scenarios, and applying appropriate mitigations—you can help to ensure the best possible experience Lync 2013 experience for all users.

2.1 Glossary WAP: Wireless access point that connects client devices (stations) to Wi-Fi. BSS: Basic service set. The basic building block of an 802.11 wireless LAN; specifically, a single

WAP with all associated stations. BSSID: The basic service set identifier. Usually, the radio Mac address(s) of the WAP. CCK: Complementary code keying. A modulation scheme used in wireless networks. Contention ratio: The ratio of maximum potential demand to network bandwidth that is actually

available. DHCP: Dynamic Host Configuration Protocol. A networking configuration protocol that is used to

automatically configure a device IP networking configuration. DSSS: Direct-sequence spread spectrum. A modulation technique where the transmitted signal is

distributed across the frequency domain, resulting in a signal with a wider bandwidth. EAP: Extensible Authentication Protocol. An authentication framework used in wireless networks. Guard interval: A time interval that ensure that separate transmissions of digital data do not

interfere with one another. See also Short guard interval. IEEE 802.11: A set of IEEE standards for the operation of wireless local area networks. OKC (also OPC): Opportunistic Key Caching. A method that reduces the handoff latency for

clients roaming between adjacent WAPs by providing preestablished authentication. Also known as Opportunistic Pairwise Master Key (PMK) Caching (OPC).

OPC. See OKC. MIMO: Multiple-input multiple-output. NAT: Network Address Translation. A process to reconfigure IP address information in IP packet

headers in a routing device. PMK: Pairwise Master Key. See also OKC and OPC. QAM: Quadrature amplitude modulation. An analog and digital amplitude modulation scheme that

uses two streams. Short guard interval: A shorter guard interval, which is possible in 802.11n. See also Guard

interval. SSID: Service Set Identifier. An association identifier between devices and Wi-Fi deployment. Wi-Fi Alliance: A global nonprofit organization that provides product certification and other

services. UNII: Unlicensed National Information Infrastructure. WMM: Wi-Fi Multimedia. A Wi-Fi Alliance specification that defines quality of service (QoS)

settings for over the air (OTA) prioritization of 802.11 frames. Also known as Wireless Multimedia Extensions (WME).

WPA2 IEEE802.11i Wi-Fi protected access: security protocol and certification program developed by Wi-Fi alliance.

WEP: Wired Equivalent Privacy (WEP) is a security algorithm for IEEE 802.11 wireless networks.

3 Usage ScenariosFor the purpose of this white paper, the Wi-Fi deployments are grouped into the following three main deployment types.

Enterprise Public hotspot Home

3.1 Enterprise Wi-FiThis deployment type is typically found in offices for professional usage of line-of-business (LOB) applications and other services provided to employees or guests. Attributes of an enterprise Wi-Fi implementation include the following:

Authentication and encryption are implemented, typically by using Wi-Fi Protected Access 2 (WPA2), enterprise wireless security software.

Multiple SSIDs are available that feature different levels of service and access, such as employee LOB SSID or guest SSID.

Multiple “thin” access points are deployed, along with Wi-Fi infrastructure controller managing WAPs.

Bandwidth and access policies may be in place.

Some small businesses may deploy a single consumer-grade WAP only, but will implement access control and security protocols. Such deployments can be considered similar to home deployments, as outlined later in this document.

3.1.1 Enterprise voice/video mobility scenariosEnterprise Wi-Fi real-time media usage can be further classified into two categories:

Fixed usage: The device typically remains stationary for the duration of a voice or video call, and maintains association with the same WAP.

Mobile usage: The user is actively moving during a call (for example, a Wi-Fi telephone device, smartphone, or tablet). The enterprise Wi-Fi and device must support a fast-handover between WAP points, with minimal impact to the audio quality, with no or only minor audio glitches.

Note that even in case of nomadic usage, devices may be redirected to a different WAP, which can occur due to active WAP load balance, or fluctuation of signal strength of adjacent WAPs.

3.1.2 Personal devices and enterprise Wi-FiAs a recent trend, enterprises are starting to allow employees to connect personal devices, such as smartphones or tablets, to the corporate Wi-Fi network.

3.2 Home Wi-FiIn home Wi-Fi deployments, an ISP provides Internet access to a single dwelling (a house, for example, or an apartment). Devices that enable local Wi-Fi are typically multipurpose devices. Attributes of a home Wi-Fi implementation include the following:

A consumer-grade wireless router connected to a cable or DSL modem. This is typically a multipurpose device that features routing, NAT, firewall, and DHCP server functionality, in addition to the wireless access point in a single unit.

Authentication and encryption are enabled, but lower security settings and key length (WEP) may be used.

WPA/WPA2 security and authentication are typically supported by all recent generation, consumer-grade WAPs.

WPS (Wi-Fi Protected Setup) support to enable easy device hookup to secured Wi-Fi home deployments.

Single access point only. Typically, a single access point is deployed to provide coverage for all home devices. However, in selected cases, a wireless repeater is deployed to increase range.

Home ISP typically implements asymmetric upload/download configurations, which produces significantly lower upstream bandwidth than downstream bandwidth.

The home Wi-Fi router provides Internet access to a wide variety of devices, such as:o Smartphoneso Tablet deviceso Game consoleso Video streaming deviceso Notebook and desktop systems

3.3 Public Wi-Fi hotspotsPublic Wi-Fi hotspots generally provide a free or a paid-per-use service, and are typically located in the following areas:

Retail locations, coffee shops, airports Hotels Airplanes, trains, buses Hospitals, public libraries

Typical attributes of public Wi-Fi hotspots:

No authentication and encryption: Wi-Fi uses the Open setting. Prior to granting Internet access, users are redirected to a Terms of Service and/or a billing page.

Access to the Internet is then granted, based on the MAC address of the wireless network adapter.

Content filtering: Black-listing filtering based on HTTP URL, or plain text search (URL and keyword filtering).

Protocol or port filtering or throttling: Specific protocol or port ranges may be blocked or throttled after a certain usage amount. Typical examples: BitTorrent and other file-sharing application port ranges.

Primarily supporting only IEEE 802.11b and 802.11g, with only few hotspots supporting IEEE 802.11n access.

Internet traffic may be channeled through a proxy, which features content filtering and/or billing and other access control mechanisms. This proxy server may not be located on-premises with the Wi-Fi deployment, which adds overall latency and affects real-time media workloads.

Asymmetric upload/download speeds with significant throttling of upstream speed, which limits the workload capacity of applications such as video chat, which require symmetrical bandwidth.

Hotspots in vehicles or remote areas can be provided through an ISP connected via a satellite or cellular data connection (adding bandwidth and latency constraints).

Due to the wide diversity of public hotspot deployments, it is difficult determine what level of service will be available. The typical goal of a public hotspot is to provide users with a basic Internet service that provides access to email, as well as web access (HTTP browser), and limits real-time or bandwidth-sensitive workloads. However, recent changes in usage patterns involving content-rich multimedia and real-time media usage make it necessary for public hotspots to provide more bandwidth and lower latency for all connected devices.

For details on Wi-Fi standards, see Appendix A: Wi-Fi Standards.

4 Issues That Affect Wi-Fi Performance for Real-Time Application

4.1 General issues

4.1.1 Wireless NIC drivers Inconsistent wireless NIC driver quality and performance. Drivers on consumer devices assume a

single WAP home scenario and show poor performance in enterprise deployments with multiple WAPs.

Background scanning delays. Depending on a driver’s strategy for updating the WAP candidate list, implementation exhibit delays of up to 1 second or more can result, causing audible glitches in real-time media applications. Ideally, the delay caused by background scanning is less than 50 milliseconds.

Drivers do not consistently use the WAP candidate list available in the beacon request. Instead, they frequently try to rescan neighbor WAPs and compile a list by themselves.

Too high or too low roaming aggressiveness. Some driver implementations are too sticky to a WAP and attempt to roam only if the signal strength has already dropped significantly and much more suitable WAPs are nearby. Conversely, drivers sometimes jump between WAPs too aggressively even when the current WAP associate provides sufficient performance. Frequent WAP handover delays affect real-time media workloads (audio glitches). Good implementations roam to a different WAP when the receive signal strength (RSS) drops below 65 dBm, and the signal-to-noise ratio (SNR) is below 30 dB.

Inconsistent TX (transmission rate) adaptation. Clients may drop too aggressively to a lower transmission rate, contributing to a congestive collapse on the WAP. Conversely, clients may stay too long at the high TX rate despite high packet loss, requiring a large number of packet retransmits.

Long WAP handover delays in the Enterprise scenarios (WPA2 Enterprise), especially observed with mobile devices. Delays are observed to be 5 seconds or more, which in most cases causes the Voice over Internet Protocol (VoIP) call or video call to be dropped.

4.1.2 Wireless NIC chipsets and hardware Inconsistent 2.4 GHz and 5 GHz support. Even in deployments with simultaneous dual-band

WAP configuration (2.4/5 GHz), as well as those that are dual-band capable, the client may default to the congested 2.4-GHz band. Wireless NIC driver default settings prefer the 2.4 GHz, but typically also offer configuration options to prefer the 5 GHz band.

Lack of antenna diversity and MIMO: Both 11g and 11n devices will benefit from multiple antennas (apart from the multistream ability of 11n), but antenna diversity is not implemented consistently and is typically not available in smartphones, due to space constraints.

Form factor restriction: Smaller devices usually have lower performance Wi-Fi radios, and smaller (mostly single) built-in antennas, which provide less signal strength than notebooks or tablet devices, which have a larger form factor. In addition, devices are held close to the body and are in motion while in use typically deliver a lower wireless performance overall.

4.1.3 Legacy interoperability issuesIn the 2.4 GHz band concurrent legacy IEEE 802.11b/g and 802.11n operation may result in interoperability issues. The presence of legacy 11b/g devices with faster 11n may adversely affect the performance of each of the device groups, or may prefer either group in accessing the physical layer (inconsistent airtime fairness). Wi-Fi equipment vendors are providing nonstandardized solutions to improve airtime fairness, such as allocating airtime dynamically for each individual client by access type, traffic, and volume.

4.2 Issues in enterprise Wi-Fi deploymentsEnterprise Wi-Fi deployments are usually fully managed, and if properly designed and configured, they can provide the best real-time media experience for mostly stationary clients. However, the following items may affect Wi-Fi performance in enterprise deployments:

1. Insufficient WAP density required to appropriately handle the bandwidth and number of concurrently connected clients.

2. Insufficient backhaul bandwidth WAPs. For example, a WAP that supports 11n by running a single 100 Mbps backhaul connection, which caps the throughout below the achievable wireless bandwidth.

3. Adjacent WAPs interference. This may result in incorrect channel selection or reduced signal strength, specifically in the 2.4-GHz band.

4. Coverage gaps. Because WAP placement is typically designed for stationary use (desktop computers) or nomadic use (notebook computers in meeting rooms), mobile users may experience coverage gaps—for example, a mobile user with a smartphone on a VoIP call in a hallway or stairwell.

5. A large meeting room with a significant number of devices that are using the same WAP may cause a high contention ratio on the physical layer and/or on the backhaul network.

6. Usage of “employee-sourced” devices. Some companies let employees connect devices that have lower-quality radios and antennas to the Wi-Fi deployment. This results in marginal signals and a compromised performance overall.

7. Support of legacy devices (such as 11b) may significantly affect the overall throughput of faster 11g and 11n devices, resulting in air time contention.

4.3 Issues in public Wi-Fi hotspotsWi-Fi hotspots are available in public locations, such as airports or hospitals, retail shops, and coffee shops to offer users Internet access. Hotspots are available either for free or for a fee that is based on time or data volume used (payload data only). For most public hotspots, no security settings are implemented. Due to legal requirements, most public hotspots let users connect to the wireless network directly only if they launch a web browser to accept usage terms or provide billing information first. Some commercial hotspots provide the options of different access classes, such as (non-real-time) data only, or real-time media, which is usually charged at a higher rate.

Typical issues in public hotspots:

Depending on the coverage area, only a single WAP may be deployed, and due to the high cost of enterprise-class WAPs, lower-grade or consumer-grade WAPs devices are typically used. The quality of the WAP and antenna configuration may negatively affect the performance.

In larger areas, such as airports or hospitals, there are typically no service-level agreements (SLAs) established, and the service provided is considered “best effort.” WAP density and coverage may be limited.

Backhaul network with limited upload speed: Most small-business Internet access packages provided by cable operators feature a relative fast download stream bandwidth, while providing only a fraction of upload bandwidth. For example, a typical package features 50 Mbps downstream and 10 Mbps upstreami. While this reasonable for typical Internet web traffic, real-time media traffic requires equal upstream and downstream performance, and so, in this scenario, only a limited number of concurrent voice/video conversations can be supported.

4.4 Issues in home Wi-Fi deploymentsWi-Fi routers deployed at home (home WAP) often have the same issue of general radio frequency noise that is present in the 2.4 GHz band. This noise is caused by neighboring Wi-Fi routers that operate on the same channel and game consoles with wireless game controllers (Bluetooth), as well as cordless phones and household appliances such as microwave ovens. Additionally, residential walls may absorb radio frequency, resulting in suboptimal signal strength.

5 Wi-Fi Deployment Recommendations5.1 Enterprise Wi-Fi To successfully implement an enterprise-grade Wi-Fi deployment that supports real-time media, it is important to perform a detailed assessment of all factors in the deployment—connected devices, usage, and prioritization of LOB wireless traffic. This section provides an outline of the planning process and describes best practices for implementing a high-performance wireless experience for all users and workloads.

5.1.1 Mixed or Wi-Fi-only enterprise deployment In most cases, the networking infrastructure in an enterprise provides wired network access, with Wi-Fi as a supplemental feature for notebooks and other mobile devices. Alternatively, enterprises may roll out Wi-Fi-only deployment for all client stations, including stationary computers. Wi-Fi-only enterprise deployment must be planned accordingly, including consideration of a high baseline usage and overall higher bandwidth and density requirements.

5.1.2 Device types in enterprise Wi-Fi deploymentsDesktop computers and notebooks are typically supplied by the employer, which makes it possible for IT departments to require standards and base specifications for the type and overall performance of the computer system, and also for the type, quality, and speed of wireless network cards.

In recent years, employees have increasingly used personal mobile devices for both personal and business purposes. In most cases, these devices have cellular and Wi-Fi capabilities, and employees naturally expect Wi-Fi connectivity. Personal mobile devices such as smartphones and tablets are most commonly used. Employees also occasionally use entertainment devices, such as game consoles in break rooms, which are typically supplied by the employer.

The IT department can deploy strategies to ensure critical LOB priority for Wi-Fi access, while providing an adequate support of personal mobile devices and entertainment devices.

One scenario for implementing multiple SSIDs and configuring for different priorities, for example, is as follows:

LOB application or production SSID (secured)i For details about the Comcast Small Business Deluxe Package that features downstream 50 Mbps and upstream 10 Mbps, see http://business.comcast.com/.

Employee personal device SSID (secured) Guest SSID (open)

5.1.3 Usage patterns in enterprise Wi-FiTo determine the scale of a Wi-Fi deployment correctly, it is necessary to assess the anticipated traffic pattern. Real-time communications and multimedia applications add significant load to the network. Specifically, to help ensure high-quality, real-time communications, VoIP and video conferencing traffic must be given higher priority than background applications, such as email.

High-density WAP deployments that feature inter-AP load-balancing can help to ensure sufficient air time for connected devices. We recommend QoS support, especially in areas with a large number of clients accessing the same set of WAPs. For details, see 5.1.5 WMM QoS and WMM-Power Save support.

For each device type, it is necessary to determine workloads that need to be supported and how the bandwidth can be distributed to meet the service level. For example, network administrators can block consumer media-streaming applications for employee personal devices, and for guest usage.

5.1.4 Windows 7 and Windows 8 considerations for Lync and SkypeApplications running on Windows 7 and Windows 8 (Classic API) have access to a set of controls to improve real-time media streaming. Settings for Windows 8 Modern applications are different and are discussed later in more detail.

Windows 7 and Windows 8 Classic (x86 only): Real-time communication applications have an option to set the network into a “streaming” mode. This setting directs the Wi-Fi NIC diver to suppress background scanning for the duration of a call and also provides improved results for stationary usage. If the user is moving during a call, however, the WAP association remains sticky until the wireless signal becomes too low and disconnects, forcing a full scan and reassociation to the nearest WAP, which usually results in an audible glitch. Note that this this streaming mode is separate from any DSCP (and resulting WMM) settings. The implementations between Lync and Skype differ in the following ways:

Lync 2013 (and Lync 2010) on Windows 7 and 8 (Classic) set this streaming mode flag (not configurable).

DSCP and WMM settings are implemented by using a QoS policy for Lync, and are usually implemented through a domain-wide policy (GPO).

Skype clients do not use the streaming mode setting.

QoS policies: Lync enables audio, video, application sharing, and filing sharing traffic to be separated to dedicated ports, which allows a port-range-based QoS policy to set different DSCP settings for each modality. Because Lync QoS policies are usually configured by using the Lync executable name, it is necessary to create a configuration for the Lync Classic (Lync.exe) and the Lync Modern experience (LyncMX.exe) on a Windows 8 x86 domain-joined computer.

Skype audio and video traffic share the same socket. Due to this limitation, it is not possible to implement different DSCP/WMM settings for audio and video traffic separately.

Windows 8 Modern (ARM and x86): The Modern API for Windows on does not expose the media streaming mode directly to applications. With the Windows Modern API, a Voice over Internet Protocol (VoIP) mode is available. If the VoIP flag is set by the application, the OS will automatically enable the streaming mode and trigger the background scanning suppression. The VoIP mode does not set any DSCP settings, and the Windows 8 Modern application does not have the ability to directly set DSCP markings.

To implement QoS settings on a standalone Windows 8 Modern computer, it is still possible to set an individual QoS policy through a custom process (scripting).

The Lync 2013 Modern experience (ARM and x86) sets the VoIP mode. Lync users on Windows 8 Modern tablet devices who are moving during a call will experience limited roaming capabilities, with the potential of handovers with audible interruptions and glitches.

Skype does not implement the VoIP mode setting on Windows 8 Modern.

5.1.5 Legacy devices in enterprise Wi-FiSupport legacy devices, such as 802.11b devices, may have a considerable impact on available air time for faster 11g and 11n devices. Network administrators may want discontinue legacy support on certain SSIDs or to deploy vendor-specific air-time fairness implementations to help ensure the prioritization of higher bandwidth endpoints over lower bandwidth legacy devices.

5.1.6 WMM QoS and WMM-Power Save support Real-time media workloads benefit from Quality of Service (QoS) implementation, in both the wired and wireless leg of the network. However, even without a QoS implementation on the wired segment, wireless QoS can benefit real-time media workloads.

WMM (Wi-Fi Multimedia), or Wireless Multimedia Extensions (WME), provides two key features, which are outlined below. While WMM-Power Save is not strictly a QoS feature, it will be discussed in the WMM context and the impact real-time media workloads on mobile devices.

WMM QoS (Quality of Service) WWM-Power Save

WMM QoS enables real-time media workloads to implement a higher QoS setting—in particular, for voice traffic over Wi-Fi. To enable support of WMM QoS end-to-end, the WAPs and client need to fully support WMM. To implement QoS on Windows Lync requires QoS policies to be configured and deployed on the server and client side, and—if applicable—on connected PSTN gateways or Session Border Controllers.

Specifically, the client OS and application level must enable the ability to insert QoS tagging, either by setting layer 3 DSCP QoS markings, which get translated to WMM settings by the wireless NIC driver, or by explicitly setting QoS/WMM tags.

For downstream traffic, the WAP must have WMM QoS enabled, and downstream traffic must carry applicable layer 3 DSCP QoS markings. In some instances, the downstream traffic does not contain DSCP markings (intermediary network elements may remove the markings). In this case, custom solutions on wireless controllers (packet inspection and heuristics) are used to identify voice traffic to enable WMM QoS, even if no DSCP markings are present downstream.

WMM-Power Save: The Power Save feature (WMM-PS) enables the client to power down the Wi-Fi radio during a sleep cycle, which is determined by interval of DTIM (Delivery Traffic Indication Message). Especially for low-power devices, such as smart phones, Power Save makes a much longer VoIP talk time possible. WWM-PS must be supported by both the WAP and the client.

Recommendation for WMM support:

The main benefit of WMM QoS appears in high-density scenarios, where a large number of clients are associated to the same WAP (for example, tradeshow floors or conferencing venues). In deployments with sufficient WAP density, and client load-balancing, there is limited benefit to WMM QoS under normal network load conditions.

WMM-PS provides the highest power benefit in scenarios where smartphones are used as VoIP phones for extended call duration (with the screen turned off). For video chat or conferencing scenarios, the power benefit is less relevant because most of the power is used by the screen backlighting. The power-saving benefit for notebook and tablet devices is also less relevant because these devices typically have much greater battery capacity.

5.1.7 Enterprise WLAN controller For enterprise Wi-Fi implementation, we recommend implementing the infrastructure with a full-featured, centralized WLAN controller and enterprise-grade managed WAPs.

The advantages of enterprise-grade WLAN controllers are as follows:

Channel configuration and signal strength are automatically adjusted to minimize radio frequency (RF) interference from adjacent WAPs based or other sources.

Support for Radio Resource Management (802.11k) to facilitate load balancing between WAPs. Support for Wireless Network Management standards (802.11v). Support for fast-handover (fast BBS transition support) implementing OKC (Opportunistic Key

Caching).

5.1.8 Recommended Wi-Fi feature support AP support of 802.11n with 3x3 MIMO support. AP support of concurrent dual-band operation (2.4 GHz and 5 GHz). Band steering support (if available) to move dual-band-capable 11n clients to 5-GHz band. Recommended WAP density for voice coverage is 3600 square feet (335 m2) with 125 percent

overlap per WAP. (WLAN equipment vendors can provide guidance for their specific equipment.) Clients need roam to a different WAP if the receive signal strength (RSS) drops below 65 dBm, or

if the signal-to-noise ratio (SNR) is below 30 dB. Higher WAP density for large conference halls and other large venues, depending on the

anticipated number of devices estimate per WAP.

5.1.9 Recommended Wi-Fi configuration settings 5 GHz band featuring 40 MHz channels to enable support of 150/300/450 Mbps 11n. 2.4 GHz band limit to 20 MHz channels to accommodate 3 non-overlapping channels, and to

mitigate RF interference (at the cost of reduced bandwidth). Depending on requirements to support multicast over Wi-Fi: Set a customized DTIM timer interval

on a separate SSID for designated to mobile devices. Implement QoS for real-time media on a backhaul-wired network (WMM (Layers 1/2, 802.1p) and

DSCP (Layer 3)).

5.1.10 Wi-Fi policies Consider separate SSIDs for personal mobile devices, with customized priority. Block legacy 802.11b access and limit legacy 802.11g access for selected SSIDs, or implement

bandwidth fairness policies (if available), based on device class/type. Implement configuration to steer dual-band-capable clients to use the 5 GHz band. Consider implementing per-device (that is, per type of device) policies. For example: Transfer

quotas for personal devices, or limitations on audio and video streaming over Wi-Fi. Policies can be based on the type of operating system that is connected to the WAP.

Recommend the use of protection of management frames.

5.1.11 Wireless network card (network adapter) recommendations Enterprise-grade notebooks that feature a high-quality 11n Wi-Fi network adapter, with dual-band

support, and at least a 2x2 configuration (multiple antenna/stream support). Desktop wireless network adapter with dual-band support 11n and at least 2x2 configuration. Mobile devices (smartphones and tablet) with full WMM support, including WMM Power Save. If possible, select dual-band capable mobile devices. Support Radio Resource Management (802.11k). Wi-Fi CERTIFIED™ 11n devices by Wi-Fi alliance (http://www.wi-fi.org). Wi-Fi CERTIFIED™ Voice Enterprise/Personal (Convergence option in certificate). Establish policies to manage network adapter driver versions, and to help ensure that Wi-Fi NIC

drivers are current. VoIP-optimized drivers that support background scanning with little or no impact on the active

connection.

5.1.12 Enterprise voice certifications Voice over Wi-Fi Enterprise Certification (VoWifi) is an emerging standard proposed by the Wi-Fi

Alliance. Certified VoWifi Wi-Fi infrastructure and clients will help to ensure to deliver best quality for real-

time media workloads.

5.2 Home Wi-Fi Home Wi-Fi deployments are typically implemented by a single multipurpose Wi-Fi router, which implements a Network Address Translation (NAT), a router, wireless WAP, and a DHCP server, and may include a cable/DSL modem functionality.

As in an enterprise Wi-Fi deployment, it may be applicable to implement multiple SSIDs on a wireless router that supports concurrent dual-band operations, and to move selected devices—such as entertainment devices and other mobile usage devices— to one SSID in the 2.4-GHz band, while running home office applications (real-time media) on a separate SSID in the 5-GHz band.

However, because the range of 5 GHz is more limited than 2.4 GHz, this involves a tradeoff between the 5 GHz band’s range and lower RF interference. In most home usage scenarios, multiple WAPs will not possible to implement.

Most home usage ISP packages deliver asymmetrical Internet access where upload bandwidth is only a fraction of the download bandwidth. To support real-time communications that use VoIP, HD video, and conferencing through a home ISP, we recommend an ISP that delivers at least 1.5 Mbps consistent uplink bandwidth. Typical routers support custom QoS for specific application traffic, based on a port range, protocol type. This prioritization is independent of the wireless leg from the device or computer to the home WAP, giving priority to real-time media traffic. Setting up custom QoS for home usage may be out of scope for most users, but it is recommended, especially if multiple devices compete for limited bandwidth.

5.2.1 Home Wi-Fi deployment recommendations Some consumer-grade WAPs support automatic channel selection (ACS), which detects RF

interference and automatically selects the channel with the least interference. If ACS is not available: Most 2.4-GHz band wireless routers are configured at channel 6 by

default, and cause interference with other WAPs in neighborhood. Changing to channel 1 or channel 11 may reduce interference.

For configuration in the 2.4 GHz band, consider using 20 MHz channel width for 11n if a significant amount of RF interferences is present. Some routers support ACS to determine how much RF interference is detected, and then select the channel width as appropriate.

If the router supports dual-band operations, consider setting the 5GHz band to use 40 MHz channel width for 11n.

Upgrade 11b and 11g wireless routers to 11n. For 11n routers, be sure that WMM is enabled (requirement to support 11n connections). 11g wireless routers may continue to be suitable to deliver real-time media traffic, depending on

concurrent usage and interference. Consider using Wi-Fi CERTIFIED™ 11n devices by Wi-Fi alliance (http://www.wi-fi.org). Consider using higher-quality network adapters with dual-band support and support for a 2x2

stream at minimum. Note that even certified 11n devices may not support 5-GHz band operation or multiple streams.

Keep Wi-Fi NIC drivers up-to-date. Enable WMM (most consumer-grade wireless routers support WMM). We recommend

implementing QoS for all devices that share the same Internet connection. Implement custom QoS policies on the WAP to prioritize traffic. For multiple device support, we

recommend concurrent dual-band wireless routers. We recommend implementing separate SSIDs to force dual-band capable devices to the 5-GHz

band because most consumer-grade WAPs do not implement the band-steering feature of

enterprise-grade IPs, and not all client devices accommodate the preference of a specific band (if dual-band capable).

Bluetooth devices, such as wireless mice, keyboards, game controllers, and cordless phones, may cause significant interference in the 2.4-GHz band. To mitigate this interference, consider wired keyboards and mice.

Avoid wireless range extenders (wireless repeaters) if possible, especially for real-time media workloads. Using repeaters may add significant latency. To mitigate range issues, relocate the wireless router or consider wireless network adapters with external high-gain antennas. For larger dwellings, consider multiple small-business-grade WAPs.

5.3 Public hotspot Wi-Fi Due to the wide variety of scenarios for deploying a public Wi-Fi hotspot, it is critical to do a complete assessment of the requirements. Public hotspots can range from the modest scale of a home usage Wi-Fi to a large, enterprise-scale deployment.

For ISP selection, consider the following factors:

ISP: Consider the ISP’s track record, availability, and whether a service level agreement (SLA) is provided.

Bandwidth and latency: Real-time media workloads require a higher upstream bandwidth and lower latency than other workloads, which typically send most of the data downstream and are less sensitive to latency.

For equipment selection, consider the following factors:

Larger public hotspot deployments typically require enterprise-grade access points and Wi-Fi controllers.

For smaller deployments, consider providing small business WAPs, which provide features beyond typical consumer-grade Wi-Fi routers.

For smaller venues with 20-100 concurrent Wi-Fi users, consider implementing multiple small business-grade WAPs that support basic load balancing.

5.3.1 Basic hotspot Wi-Fi configurationsSmall hotspot: Up to 15 concurrent Wi-Fi users.

Single small business-grade WAP 2.4 GHz with 3x3 antenna configuration Per device bandwidth throttling (if implemented)

Medium hotspot: 15-100 concurrent Wi-Fi users.

3 small business-grade 2.4 GHz with 3x3 antenna configuration Same SSID of all WAPs Spread channels : 1,6, 11 Per device bandwidth throttling (if implemented)

Large hotspot: 100+ concurrent Wi-Fi users.

Entry-level enterprise WLAN controller 4+ entry-level enterprise WAPs (single band) 2.4 GHz Per-device bandwidth throttling

Appendix

1 Appendix A: Wi-Fi Standards The following section provides an overview of existing and future Wi-Fi standards and their impacts on real-time media quality.

1.1 IEEE 802.11a/b/g/n Wi-Fi standardsMultiple wireless Wi-Fi standards are based on the original Wi-Fi 802.11-1997 working group deliverable. The following table provides an overview of the Wi-Fi standards that have been adopted by typical enterprise, retail/hotspot, and home Wi-Fi implementations.

Wi-Fi Standard

GHz Raw Mbps

Comment

802.11a 5 54 Common in enterprise deployments.

802.11b 2.4 11 First widely adopted standard in consumer and enterprise space.

802.11g 2.4 54 Widely adopted standard succeeding 802.11b.

802.11n 2.4, 5 54 - 600 Currently fasted widely adopted Wi-Fi standard. Typically 150-300 Mbps.

1.2 802.11a (legacy standards)The original 11a standard has not been broadly adopted, mostly due to regulatory issues (usage 5GHz band) and technology issues with first-generation devices. From a technical perspective, however, 11a was technically superior to the more widely adopted 11b. Devices that use the second generation of the 11a standard have been more widely adopted, even though 11a can be considered a legacy standard.

11a allows for a total of 12 non-overlapping channels (20 MHz bandwidth) in the 5-GHz band with a maximum raw data rate of 54 Mbps. Data rates can be reduced to 6 Mbps if signal degradation is detected.

1.3 802.11b (legacy standards)The 11b standard has been broadly adopted by enterprise and home Wi-Fi deployments. It has a raw data rate of 11 Mbps, uses the unlicensed 2.4 GHz band (DSSS CCK), and supports an adaptive rate down to 1 Mbps. Although 13 channels are defined in the 2.4 GHz band, most channels have significant overlap. Counting only non-overlapping channels, the total available number of channels is only 3 (Channels 1, 6, 11). Certain countries—such as Japan—allow 14 channels, which enables four non-overlapping channels total. Although 11b had been widely adopted, it has been mostly replaced by the more recent 11g deployments.

The typical indoor range is about 30 meters (100 feet) at full speed, and about up to 90 meters (300 feet) at 1 Mbps.

Bandwidth provided by 11b can support real-time media streaming audio (VoIP) and video for a single user connected to a WAP (such as in a home usage scenario). However, net available bandwidth may drop below levels for sufficient throughput to support high-definition (720p) video. Multiple 11b devices that share the same WAP typically will not have sufficient bandwidth to support high-quality, real-time media stream.

1.4 802.11g (current standards) The 11g standard replaced the widely deployed 11b standard, increasing the raw data rate up to 54 Mbps in the 2.4 GHz band. All 11g devices are fully backward compatible with 11b, and it is possible to operate 11b and 11g devices in the same channel. However, concurrent access for 11b and 11g devices reduces throughput for the faster 11g devices. Like 11b, the 11g standard uses the 2.4 GHz band, with similar limitations. A total of four non-overlapping channels are available.

Data rates dynamically adjust down to 1 Mbps.

Using multiple antennas on devices and WAPs provides a better signal, which mitigates multipath fading effects: transmission signals reflect on walls and objects, and distort signals at the receiver. Note that devices with multiple antennas for 11g do not implement MIMO, which is only available with 11n.

Bandwidth provided by 11g can support real-time media streaming audio (VoIP) and high-definition video, but bandwidth limits may be reached if multiple devices access the same WAP.

1.5 802.11n (current standards)The 11n standard provides a big step up in available throughput, which accommodates data rates from 54 Mbps up to 600 Mbps by using multiple simultaneous streams and currently represents the fastest Wi-Fi standard. Typically, 11n implementations support two streams, for a total of 300 Mbps.

11n devices, such as 11b/g, operate in the 2.4 GHz, but also extend to the 5 GHz band, such as 11a. A wider 40 MHz frequency band option has been added to enable higher throughput.

Because 11n provides more multiple different configuration choices, it is critical for administrators in charge of high-performance deployment to understand all the technical implications, especially when 11n is deployed with concurrent 11b/g access in the 2.4 GHz band. While 11n was finally certified by the Wi-Fi alliance in 2009, multiple 11n draft implementations had been commercially been available prior to setting the final standard.

Bandwidth provided by 11n can support all real-time media streaming audio (VoIP) and high-definition video. WAP congestion can occur with 11n, especially if legacy 11b/g devices are used concurrently in the 2.4 GHz band.

1.5.1 802.11n MIMOMIMO is part of the 11n standard and enables the sender and the receiver to use multiple antennas to perform Spatial Division Multiplexing (SDM). A single data load can be transferred as multiple multiplexed data streams within the same channel. Using multiple data streams concurrently increases the maximum available aggregated bandwidth. To achieve MIMO, however, both sender and receiver must provide multiple discrete antennas and multiple radios and Analog-Digital (AD) converters for each stream, which increases the cost and power requirements per device. MIMO is not required for 11n certification, and is an optional feature.

To identify devices and their respective MIMO capabilities, the following notation is T X R: S, where T represents the total number or transmitting antennas, R represents the total number of receiving antennas, and S indicates the total number of separate spatial streams. The following table provides an overview of MIMO configurations:

11n MIMO

Raw data rate (Mbps)*

Comments

1 X 1: 1 150 Single antenna 11n basic implementation. Typically found in mobile devices, such as smartphones and basic tablet computers, as well as USB network adaptors and basic Wi-Fi desktop interface cards.

2 X 2: 2 300 Basic 300 Mbps option with dual send-and-receive antennas. Typically found in notebook computers and mid-level Wi-Fi desktop interface cards. No antenna

diversity while using two streams.

2 X 3: 2 300 Improved antenna diversity and improved range. Typically found in high-quality notebook and Wi-Fi desktop interface cards.

3 X 3: 2 300 Further improved antenna diversity and improved range.

3 X 3: 3 450 Support for three concurrent streams to support 450 Mbps. Found in only a few commercially available devices and WAPs.

4 X 4: 4 600 Maximum data rate available with the 11n standard, using four concurrent streams. No commercially available devices yet.

* Maximum raw data rate that uses a 40-MHz channel, which uses the short guard interval (SGI).

1.5.2 802.11n Spatial diversityIf limited connectivity is detected, the same data stream can be transmitted by using multiple spatial streams. At the receiver, the data is combined to extract a single original stream. This adds robustness and reduces the number of retransmit requests under poor condition. To support this feature, the sender and the receiver are required to have at least two send antennas and two receive antennas.

1.5.3 Physical layer access by using a 40-MHz wide bandAnother improvement of the 11n standard is to define a 40-MHz wide band (compared to the 20 MHz available with 11a/b/g) to enable higher bandwidth. However, using 40 MHz in the 2.4-GHz band allocates two non-overlapping channels, which further reduces the number of available channels and poses coexistence challenges if legacy devices (11b/g) operate in the 2.4-GHz band concurrently with 11n devices that use 40-MHz wide channels. Implementing 40 MHz bandwidth in the 5 GHz is typically not problematic because there are more channels available, and less RF interference is observed.

1.5.4 Wi-Fi Certified 802.11 n devices The 11n standard establishes a number of additional features to enhance performance and robustness. The Wi-Fi Alliance established a certification program for devices, WI-FI CERTIFIED™ n ii, which defines a baseline of supported features. As part of the certificate for 11n, the supported features are listed:

Spatial streams: Devices must support the transmit and receive function of at least two spatial streams.

Block ACK protocol: Sends a single block acknowledgement (ACK) frame to acknowledge several received frames.

Short guard interval (SGI): Short GI is 400 nanoseconds vs. the traditional GI of 800 nanoseconds (10 percent improvement). For example, using SGI reduces the maximum data rate from 150 Mbps to 135 Mbps per stream.

Greenfield preamble: A technique that enables an 802.11n network to use a shorter preamble to improve the efficiency and power consumption of 802.11n networks. Recommended if no legacy devices are present (for example, in the 5 GHz band reserved for 11n devices, and if no 11a devices are present).

Space-time block coding (STBC): Improves reception by coding the data stream in blocks that are distributed for transmission across multiple transmitting antennas and across time.

A-MPDU: Aggregates MPDUs (MAC protocol data units) to include more information in each exchange and reduce the header and the interframe gap overhead at the MAC layer.

Aggregation Protocols in receive modes A-MPDU and A-MSDU (MAC service data unit): Sends a single block acknowledgement (ACK) frame to acknowledge several received frames.

WMM Power Save support enables low-powered devices to stay in sleep mode longer, therefore significantly reducing the power requirements. WAP must support WMM Power Save, and a suitable DTIM interval must be determined as a compromise between latency and battery saving.

ii For details about Wi-Fi CERTIFIED™ n – Certification for Wi-Fi devices, see http://www.wi-fi.org/certification/programs.

5 GHz support: The 11n standard allows for 2.4 GHz and 5 GHz. However, it does not also require devices to support 5 GHz. As a result, many 11n devices provide support for only the 2.4 GHz band. Typical enterprise-grade notebook support is dual-band 2.4 GHz and 5 GHz, and it should be configured for the 5 GHz band.

1.6 802.22 and 802.11af Wireless Regional Network (future standards)The IEEE 802.22 draft standard—also known as “Super” Wi-Fi, or “whitespace” Wi-Fi—is a standard to enable Wireless Regional Area Network (WRAN) standard. The operating frequency is in the “white space” left behind by the unused TV broadcast spectrum that became available with the transition to digital TV. This spectrum is between 54 MHz and 862 MHz, and is shared with TV channels (channel usage is variable by region).

Due to the much lower frequency band, a typical 802.22 base station would be able to cover devices in a radius of up to 20 miles, at a rate of approx. 20 Mbps.

The 802.22 standard proposes different techniques (cognitive radio techniques), spectrum-sensing, and different modulation techniques, depending on the distance of a client from a base station. Cognitive radio techniques enable a transceiver to automatically adjust transmission and reception parameters based on locally available channels, and level of radio interference detected, as part of the overall dynamic spectrum sensing process.

802.11af is a similar effort to 802.22, driven by the Wi-Fi Alliance. This draft standard also uses cognitive radio techniques and geographical data to determine available TV whitespace frequencies.

Neither the 802.22 nor the 802.11af standard is currently ratified, and, except for pilot deployments, no commercial devices are available at this time. Therefore, it is too early to assess how well 802.22 or 802.11af will work regarding real-time media traffic.

1.7 802.11ac Gigabit Wi-Fi (future standards)802.11ac, or Gigabit Wi-Fi, is a standard currently under development. This standard will enable single-link throughput of at least 433 Mbps. This standard extends concepts of 11n in the following areas:

Up to 8 multistreams (MIMO), up from a maximum of 4 with 11n Wider bandwidth with a minimum of 80 and a maximum of 160 MHz, up from 20 to 40 MHz with

11n Multiuser MIMO (multiple radios) High density modulation up to 256 QAM, up from 64 QAM 60-GHz band for specialized application

The following table shows anticipated usage profiles and bandwidth:

Device type AP and device configuration Channel width Total bandwidth

Smartphone 1 antenna WAP, 1 antenna station 80 MHz 433 Mbps

Tablet 1 antenna WAP, 1 antenna station 160 MHz 867 Mbps

Notebook 2 antenna WAP, 2 antenna station 160 MHz 1.73 Gbps

TV set-top box, Digital TV

8 antenna WAP, 8 antenna station 160 MHz 6.93 Gbps

802.11ac will provide sufficient bandwidth for all real-time media applications. The standard is designed to enable quality similar to HDMI throughput. Raw HDMI 1.4 maximum bandwidth is 10.2 Gbps.

Standards and vendors: Wireless Gigabit Alliance (http://www.wigig.org) is driving the specification and standardization process, together with the Wi-Fi Alliance.

Gigabit Wi-Fi should work well for real-time media traffic. However, significant range restrictions between base station and client are expected, which will restrict highly mobile usage.

2 Appendix B: Wi-Fi Frequencies 2.1 2.4-GHz bandAlthough the initial 11a standard was limited to 5 GHz, a breakthrough for local area wireless in consumer networking came with the 11b and 11g standards in the 2.4 GHz band. This was, in part, because of the cheaper and more readily available devices, as well as the usage of the unlicensed 2.4-GHz band. In addition, the 2.4-GHz band is available in most countries.

However, operating the in the 2.4-GHz band is problematic because this unlicensed frequency range is shared by other devices, such as Bluetooth devices, cordless phones, baby monitors, and some kitchen appliances. As a result, especially in high-density residential areas (such as apartment buildings), interference with in the 2.4-GHz band is very common. In addition, the 2.4-GHz band (2.4000–2.4835 GHz) is divided into multiple overlapping channels with 20-MHz channel width, which limits the total number of available non-overlapping channels to typically only three (11b) or four (11g/n) channels. Furthermore, using 11n standard on the 2.4 GHz and using the wider 40-MHz channel width further limits the number of available channels.

The follow figure provides an overview of non-overlapping Wi-Fi channel allocation in the 2.4-GHz band for 11b and 11g/n.

2.1.1 Coexistence of 11b/g/n in 2.4 GHzBecause most Wi-Fi deployments require backward compatibility to 11b/g (or at least 11g support), faster 11n devices are affected due to the presence of other 11b/g devices on the same channel. In this case,

the airtime is shared with slower devices that occupy the physical layer for a longer time, affecting faster 11n devices. Some Wi-Fi equipment vendors offer custom (nonstandardized) solutions to address this issue, to help ensure that faster 11n devices get a fair share of air time.

2.2 5 GHz band The unlicensed 5 GHz is band has generally much less interference compared to the crowded 2.4 GHz band, and has also an overall wider spectrum to enable a larger number of non-overlapping channels. A typical enterprise-grade WAP will support 21 channels with 20 MHz bandwidth or 8 channels with 40 MHz bandwidth. A limiting factor, however, is the physical fact that the higher the frequency of a wireless signal, the shorter its range. In particular, the higher frequency wireless signals of 5 GHz networks do not penetrate walls and other solid objects nearly as well as do 2.4 GHz signals, therefore limiting their reach. This disadvantage is, in part, offset by less RF interference. Channels in the 5 GHz band are not continuously available, and they are split into a number of sections (UNII bands). In addition, there are additional limitations around channel availability based on country regulations and other local regulations. It is important to understand prior to implementation what settings are permissible, and what process for deployment is required by local authorities.

2.3 Quality of Service and Wi-Fi Multimedia (WMM)Wi-Fi Multimedia (WMM), also referred to as Wireless Multimedia Extensions (WME), enables basic Quality of Service (QoS) based IEEE 802.11e.

WMM replaces the Wi-Fi DCF distributed coordination function for CSMA/CA wireless frame transmission with EDCF (Enhanced Distributed Coordination Function). WMM sets the Transmission Opportunity (TXOP) to a shorter time window for video and voice traffic, compared to regular data without QoS requirements.

WWM provides four priorities, as shown in the following table:

Priority description

ID Max TXOP Comment

Background AC_BK Default Lowest priority (deferred)

Best effort AC_BE Default Regular

Video AC_VI 3.008 msec Video traffic

Voice AC_VO 1.504 msec Voice/VoIP (highest) priority

2.3.1 WMM prerequisites Client to Access Point direction: To implement higher WMM priorities, the client device and the

VoIP/video application, or the client operating system, must set a QoS setting (for example, a DSCP tag) correctly. Some mobile devices may not support this on the NIC driver, operating system, or application level correctly. Windows 7 and Windows Server 2008 operating system support fully customizable QoS policies.

Return data sent from a server, such as the Lync Conferencing Server, must be tagged for higher QoS tags in order for WAPs to implement WMM priorities correctly. This is required, even if the wired (backhaul) network does not implement QoS correctly. Note that data packets received by the WAP may originate from servers on the Internet and may have their DSCP tags removed on intermediate routers or gateways.

WMM affects only the wireless client connection to the WAP. Further QoS configuration for wired LAN should to be implemented on the backhaul segment to ensure that voice and video traffic gets preference there as well, although this is not a requirement.

Custom solutions implemented by wireless equipment manufacturers perform packet inspection and packet retagging to implement the proper WMM level. For example, payloads containing an audio codec (G.722), or a video codec (H.264) are identified and prioritized. Note that this package inspection can prioritize downstream traffic only to a client device. For upstream traffic, the client device must be aware of increased WMM priority (that is, setting a shorter Transmission Opportunity Interval (TXOP)).

2.3.2 WMM-Power Save (APSD)WMM-Power Save, or automatic power save delivery (APSD) is part of the 802.11e specification. WAPSD provides a power management mechanism for mobile devices, specifically around voice and video real-time data. This mode enables mobile devices to perform a more efficient usage of the wireless radio, with reduced battery impact. Device and WAP sync on a specific time interval, and WAPs buffer downstream data until the device powers up the wireless radio, on a defined interval. This enables the send-and-receive transfer to occur in the same “awake” window, and makes it possible for the device to remain in a low power mode longer. (For more details, see 8.5 WMM-Power Save and DTIM interval.)

3 Appendix C: Wi-Fi Security Most Wi-Fi deployments implement encryption and/or authentication for connected wireless devices, with the exception of Wi-Fi deployed in hotspots or for guests. Authentication and encryption are adding overhead to Wi-Fi communication, and especially with real-time media workloads, it is important to mitigate any delay. This chapter discusses Wi-Fi security standards and implications to real-time media, especially for mobile users roaming between multiple WAPs during a call.

The following figure summarizes the three primary security settings for Wi-Fi deployment:

Security setting Description Comment

WEP Wired Equivalent Privacy WEP uses a single encryption key for all devices and all packets, and has minimal overhead.

WPA Wi-Fi Protected Access WPA includes stronger encryption, key management, and authentication mechanisms, and adds more overhead.

WPA2 Wi-Fi Protected Access 2 Based on draft 3.0 of the 802.11i standard, adding Advanced Encryption Standard (AES) to WPA. WPA2 options are Personal or Enterprise.

The security level needs to be mutually supported by the WAP and the device. Some older devices may not support WPA or WPA2, or may have incompatibly implementations for WPA and WPA2. Typically, WEP represents the lowest common denominator in Wi-Fi security and is supported by all devices, including legacy devices.

In addition to the security protocol, it is possible increase encryption complexity by configuring longer keys, which typically adds more overhead and requires more processing power on devices. This also adds a small amount of latency.

3.1 Wi-Fi security and real-time media workloadsThe WPA2 (IEEE 802.11i-2004) security mechanism provides the most secure authentication method in Wi-Fi deployments. It is the currently established standard in enterprise deployments.

WPA2 supports two different authentication modes:

Personal: Personal mode uses a password-based authentication through preshared Key (PSK). This mode does not require reauthentication if a user roams from one WAP to another WAP; however, the administrative overhead and security limitations make the Personal mode not suitable for most enterprise deployments.

Enterprise: Enterprise mode requires wireless devices to mutually validate credentials with a backend authentication server by using Extensible Authentication Protocol (EAP) or RADIUS 802.1x. The Enterprise WPA2 is the standard, and also the best practice security standard, implemented in enterprise deployments.

Although WPA2 is most secure, it adds authentication overhead (delay) for clients who roam between WAPs. Specifically, if a client roams between WAPs while in a VoIP (VoIP/video) call, the necessary authentication overhead causes a temporary interruption in the connection, which then causes a glitch in the ongoing session, or, in the worst case, ends the session completely.

To mitigate this handover delay and keep it below a threshold of 50 milliseconds (ms), several methods have been implemented. (Within 50 ms delay, there is typically no or very little impact on an existing real-time communications session.)

Solutions to reduce handover delays are called fast handoff or fast transition solutions. The following section provides technical details about the authentication message exchange and information about how the delays can be reduced.

3.1.1 Pairwise Master Key Security Association (PMKSA)WPA2 defines an authentication method between a device and WAP. After successful authentication, a Pairwise Master Key (PMK; also called Master Session Key, or MSK) in the EAP is generated by an authentication server and delivered to the WAP, where it is stored. If a client now roams to an adjacent WAP, the WAP determines whether a valid PMK is present, and, if not, the WAP performs another authentication request by using EAP. Depending on the infrastructure deployment, and load on the authentication server, the end-to-end time of this reauthentication is longer than the acceptable threshold of 50 ms.

3.1.2 Opportunistic Pairwise Master Key (PMK) CachingOpportunistic Pairwise Master Key (PMK) Caching (OCP; also known as OKC, for Opportunistic Key Caching) is a method that reduces the delay needed for reauthentication. This method reduces the handoff latency for clients roaming between adjacent WAPs by providing a preestablished authentication. WAPs are grouped into mobility groups (also known as mobility zones) controlled through a central Wi-Fi controller. If a client enters a mobility zone, a PMK is generated and then automatically distributed to all WAPs in this zone. This proactive distribution of the PMK is performed by a centralized Wi-Fi controller. When the key is present, and the client associates to the WAP, the authentication delay is significantly reduced.

Opportunistic Pairwise Master Key (PMK) Caching is not standardized by the IEEE. Nevertheless, it is supported by many wireless network vendors. Standardization is completed through IEEE 802.11r-2008, which is described in the following section.

3.1.3 802.11r fast BSS transitionIEEE 802.11r-2008 or fast BSS (Basic Service Set) transition is standard, and is designed to address the handover delay in a way similar to Opportunistic Pairwise Master Key (PMK) Caching. The main differences are that 802.11r specifies the key hierarchy in more detail and also reestablishes QoS admission control. Although not widely available in commercial products, 802.11r is expected to be adopted by all wireless network vendors for future products or firmware updates.

The objective of 802.11r is to explicitly address fast handover of mobile devices that run VoIP and other real-time media sessions.

4 Appendix D: Wi-Fi WAP Handover and Resource Management

4.1 Background WAP scanning When Wi-Fi devices are associated to a WAP, they must periodically perform a rescan to update the list of available nearby WAPs. Scanning is performed in the background, but can affect the existing data connection to the WAP, and—depending on scanning method and algorithm used—may compromise an ongoing real-time media connection.

Devices have the following options to determine adjacent WAPs:

Passive scanning: A device listens per channel for the WAP’s beacon. In this case, the delay is typically up to 100 ms (up to the maximum beacon interval).

Active scanning: A device issues an active probe request and waits for probe response from the WAP. Active scans are performed if a device has no current connection, or has lost the connection. The delay here is typically shorter (approximately 20 ms).

Interleaved scanning: To minimize impact and reduce delays, a device scans only a single channel at a time and continues regular transmission between scans. In contrast, noninterleaved scanning halts the current transmission and scans all channels, which generates a longer delay and adds jitter.

4.2 802.11k Radio resource management802.11k (radio resource management) improves the load balancing of traffic in a Wi-Fi deployment. Typically, a device will connect to the WAP that provides the strongest signal, without factoring in the current number of connected devices or the load on the WAP. 802.11k makes it possible for the device to locate and connect to an optimal WAP, including the considerations of load and signal strength. 802.11k is especially efficient in large-venue deployments, such as large conference rooms with multiple adjacent WAPs. The 802.11k and 802.11r standards are further extended in the more recent 802.11v standard. (For details, see 8.4 802.11v Wireless network management.)

Site table delivery: This feature enables the Wi-Fi infrastructure to deliver a list of available WAPs to devices, which removes the need for a client to obtain and maintain a WAP site table by frequently scanning for available WAPs. Both the WAP and the device must support 802.11k to enable site table delivery.

AP beacon and WAP neighbor reports: Devices can implement 802.11k features to support receiving and interpreting WAP beacon and available neighbor WAP reports. This improves WAP load-balancing and roaming support, which, in turn, improves overall real-time media performance. The neighbor report provides devices with information about adjacent WAPs to optimize roaming decisions in the client.

In summary, to minimize impact to any existing data connection, there are different strategy and algorithms available, and the resulting effectiveness of the scanning process depends on the NIC chipset, the driver, and the operating system on the device, as well as the features available on the infrastructure.

Wireless equipment vendors have implemented custom solutions outside the 802.11k standard that address the same needs, but do not require any special device support.

4.3 Other industry-standard fast handover solutionsThe Cisco Compatible Extensions (CCX) V3—a proprietary wireless LAN protocols extension—enables fast transitions on supported devices. Part of CCX, the Lightweight Extensible Authentication Protocol (LEAP) predates the 802.11i IEEE standardization, and due to security concerns, Cisco now recommends the use of LEAP-FAST (Flexible Authentication via Secure Tunneling). Implementations of CCX require both the WAP and the device to be CCX-certified (NIC and drivers).

4.4 802.11v Wireless network management802.11v, which was ratified in February 2011, defines several functions in different Wi-Fi management that go beyond specific requirements to support real-time media over Wi-Fi.

Following are the areas addressed with 802.11v:

Power management: Extends the power savings so that it’s possible to power down radios for longer duration.

Device and WAP management: 11v BSS transition management enables WAPs to exchange more networking topology information with devices to allow for better WAP load balancing (beyond 802.11k).

Location and time sync: Timing synchronization enables accurate device location and facilitates media broadcast audio synchronization (Wi-Fi speakers). Location synchronization enables support for Wi-Fi RFID tags and for implementing a more precise emergency location for enterprise voice-integrated devices (for example, Enhanced 9-1-1 (E9-1-1)).

Other 11v specifications:

Enables devices to send certain types of events for diagnostics, and enables WAPs to collect metrics to improve application performance.

Interference reporting and mitigation. Support for authentication troubleshooting.

4.5 WMM-Power Save and DTIM intervalWAPs generate beacon frames to include a Traffic Indication Map (TIM) to advertised clients of buffered frames at the WAP, targeting individual clients (unicast). Clients then request buffered frames from the WAP through a query, and WAPs deliver the buffered frames to each client.

This polling approach, however, is not suitable for broadcast and multicast frames, because the WAP would have to transmit frames to each client individually. For broadcast and multicast scenarios, the frames are delivered at a set time indicated by a DTIM (Delivery Traffic Indication Message) interval, and all clients will listen for incoming frames in this interval. (The DTIM interval is a parameter that is set by the Wi-Fi infrastructure, and advertised by each WAP to all clients associated to that particular WAP at a given time.

Mobile clients are now required to power up their Wi-Fi radios at the interval advertised through DTIM in order to avoid missing any multicast or broadcast frames. Increasing the DTIM interval will enable mobile clients to remain longer in the low-powered mode. However, a longer DTIM interval in multicast/broadcast applications means that the WAPs are required to buffer more data and to observe higher latency to deliver these frames to clients. Before increasing the DTIM interval, it is necessary to assess whether multicast/broadcast is widely used—for example, in real-time multimedia broadcasting. Typical adjustments for settings include, for example, changing the DTIM default of 100 ms to 200 ms, or to 300 ms.

Wi-Fi vendors typically provide best practices for setting the DTIM in the Wi-Fi infrastructure for a number of different profiles and anticipated usage. Unicast traffic is not affected by DTIM settings. We recommend a dedicated SSID for mobile devices.